PROJECT SUMMARY This research is for the development of new approaches to the analysis of data from large cohort studies, either epidemiologic or clinical trials, with many qualitatively different variables observed over several time points, with exploratory genetic marker data vectors of length the order of thousands, with pedigree information, and with multiple correlated outcomes of interest. The aim is to develop methods for discovering relevant components or patterns of components in long or ultra-long genetic marker data vectors as they interact in conjunction with patterns or clusters of clinical/lifestyle/environmental variables and, possibly, pedigrees to suggest unusually high risk and correlations of interest for multi- ple outcomes of interest. The methods proposed are an attempt to develop tools that merge so-called data mining approaches and penalized likelihood methods (some developed in prior grants) and that have the ability to generate hypotheses which can then be examined more closely by classical para- metric statistical methods and by experimenters to formulate further hypotheses. The emphasis is on approaches that have the ability to contribute information to evidence-based and personalized medical decision making. Data from Beaver Dam Eye Study will be used to examine the models under study for their reasonableness and for their ability to answer questions meaningful to the study scientists. The results will have broad applicability to other large epidemiological studies as well as to clinical tri- als, in particular those collecting local and genome-wide genetic marker information along with other, heterogenous risk factors.